PD-1-Mediated T Cell Exhaustion Is Prevalent Among Patients with MPN-Associated Myelofibrosis Independent of JAK1/2 Inhibition

Background: Primary myelofibrosis (PMF), post-polycythemia vera MF (post-PV MF) and post-essential thrombocythemia MF (post-ET MF) are characterized by expansion of the neoplastic clone and by progressive bone marrow (BM) fibrosis. Like in other hematologic malignancies, in most patients with MF the immune system is significantly deregulated: MF patients' plasma cytokine and chemokine levels are markedly increased and their normal T cell subset distribution is significantly altered. Although treatment with the Janus kinase (JAK)-1/2 inhibitor ruxolitinib significantly decreases cytokine/chemokine levels, reduces spleen size, and improves symptoms and quality of life, it does not reverse BM fibrosis nor does it halt the propagation of the neoplastic clone. The T cell immune checkpoint programmed cell death protein-1 (PD-1) promotes immune tolerance by binding to the tumor's cell surface PD-1 ligand (PD-L1). Whereas the importance of T cell-mediated immune tolerance in MF has been documented and trials evaluating clinical benefits of PD1/PD-L1 checkpoint inhibition are ongoing, little is known about the effect of ruxolitinib on PD-1 expression in T cell subsets. Therefore we systematically analyzed MF patients circulating T cells' surface marker expression prior to and during ruxolitinib treatment.

Methods: Peripheral blood cells were obtained from well-characterized PMF, post-PV MF and post-ET MF patients prior to and during the course of treatment with ruxolitinib (n=47) and, as control, from age-matched healthy donors (n=28). The proportion of PD-1-expressing CD4⁺ and CD8⁺ cells was assessed using multiparameter flow cytometry. Naïve, central memory, effector memory, and effector T cell subsets were defined based on CD45RO and CD27 cell surface antigen expression.

Results: A significantly high number of circulating T cells co-expressing CD4⁺/PD-1⁺ and CD8⁺/PD-1⁺ was found in MF patients compared to age-matched healthy individuals (5.3±4.1% vs. 3.4±1.7%, P=0.028; 7.1±4.4% vs. 3.8±2.3%, P=0.001). Whereas MF patients' naïve T cells harbored an increased number of cells co-expressing CD8⁺/PD-1⁺ (P=0.007), but not CD4⁺/PD-1⁺, their T central memory cells had a high proportion of cells co-expressing CD4⁺/PD-1⁺ and CD8⁺/PD-1⁺ (P<0.001; P<0.001). Similarly, a high proportion of circulating PD-1⁺ T effector memory cells (P<0.001; P<0.001), and T effector cells (P=0.013; P<0.001) was found in MF patients compared to the same cell subsets in healthy age-matched individuals. The proportions of PD-1⁺ T cells significantly correlated with LDH level and DIPSS score (CD4⁺ T cells), monocyte count (CD8⁺ T cells), and total leukocyte count and spleen size (both subsets). Remarkably, the percentage of PD-1⁺ cells within naïve and central memory CD8⁺ T cell populations was significantly higher in MF patients with circulating blasts (P=0.036). To determine the effects of ruxolitinib administration, we performed repeated flow cytometry analyses on MF patients' T cells prior to and during treatment (median duration: 4.3 years). Overall, no significant change in PD-1 expression levels in any of the different T cell subsets was detected over the entire treatment period. However, a significant reduction in percentage of cells co-expressing CD4⁺/PD-1⁺ and CD8⁺/ PD-1⁺ compared to treatment baseline (4.4±0.4% vs. 7.6±2.0%, P=0.011; 6.3±0.6% vs. 10.4±2.7%, P=0.021) was found in patients whose spleen size was reduced after 6 months of treatment.

Conclusions: In patients with MF, circulating T cells express high levels of PD-1. While not restricted to a particular stage of T cell differentiation, the correlation between PD-1-expressing T cells and distinct clinical parameters suggests that increased PD-1 levels might induce immune exhaustion in T cell subsets in different ways. Although ruxolitinib significantly inhibits the JAK1/2 signaling pathway in a variety of hematopoietic cells, thereby lowering cytokine/chemokine levels in almost all MF patients, treatment with ruxolitinib did not affect PD-1 expression nor did it alter its distribution among the T cell subsets. Yet, the proportion of PD-1-expressing CD4⁺ and CD8⁺ cells was markedly reduced in patients who experienced a superior response to ruxolitinib as assessed by significant spleen size reduction. How disease burden and MF microenvironment affect PD-1 expression in T cells of patients with MF warrants further investigation.